9 November 2018

Why R?

Preaching to the choir

Why R?

  • Reproducible work (often much quicker)
  • Quality assurance
  • Work is easily updated
  • Works well with report writing processes
  • Easy movement of data into/out of other formats
  • Useable graphic outputs

tRansition issues

Fear of change

tRansition issues

  • Legacy programs
  • Existing skills
  • Training
  • Skill development

What even is a workflow?

What is involved in each of these steps?

What barriers are there to using R?

  • Minimise repetition; scope for error; unneccessary work
  • Personal preferences vs Outputs

Supportive environment is critical

Computing

  • Access to R Studio
  • Access to packages, internet proxies
  • Version control software

Work

  • Culture of learning
  • R user group ('Community of interest')
    • Webpage
    • Regular interactive sessions
    • Sharing expertise between teams

Supportive environment is critical

In order to achieve this, you need to sell the benefits

  • Reproducible, auditable work

  • Open source software

  • Efficiency gains

  • Knowledge sharing

Tidyverse

  • Changed how data is imported, tidied, transformed, visualised
  • Common standard for data manipulation
  • Human 'readable' code
table <- SurveyFrame %>%
   # Group by two factor variables
   group_by(Response, Profit.status) %>%                           
   # Summarise 
   summarise(`Total assets` = sum(Total_assets, na.rm=TRUE),       
             `Accounts` = sum(N_Accounts, na.rm=TRUE)/1000,
             `Number of funds` = n() ) %>%
   # Turn into percentages
   mutate_if(is.numeric, funs(`(%)` = round(./sum(.)*100, 2) ) )

R Studio

  • Ease of managing R sessions, accessing files, outputs

Projects!

  • One folder/project

  • Transferable work

  • No more setwd

  • Use here package to build portable paths within projects

  • Multiple projects at once

Version control

  • Access git from within R Studio

  • Link with github/gitlab

  • Requires support of IT

  • Ease of sharing code/collaborating

Style management

  • Fundamental to integration of R with other work

  • PCcharts package produces standard graphics with ggplot

R Markdown

  • Reproducible record of work within R
  • Outputs accompanied by text
  • Report in a range of formats
    • html; docx; pdf; ioslides
  • Share R outputs with non-R users
  • Easy revision/updates

We will have a closer look at using R Markdown today

Interactive data exploration

  • Better understanding of raw data
  • Tool for cleaning
  • Means of sharing complex data

Different approaches:

  • Shiny
  • plotly